PromptLabX
PromptLabXPrompt engineering practice

The best engineers write the best prompts.

AI hasn’t replaced engineers. It has raised the bar. Practise on deliberately vague problems and get scored on how well you think, not just what you ask.

Try a problemSee top attempts

132 problems across system design and debugging. Free to use.

The app is too slow.Debugging

Goal: explore requirements

Your prompt

A few guesses: the frontend is probably re-rendering too much, the API gateway might be hitting a distant region for some users, and the queries likely aren’t indexed. I’d start with the query layer.

AI response

Good instinct to narrow it down. It’s slow for everyone on the main feed, and yes, the database is the culprit. Queries aren’t indexed and the feed hits it on every load…

Prompt score
8.2 / 10

Code is no longer the bottleneck.

For decades, writing code was the scarce skill. Now AI writes the code. What it can’t do is think clearly about the problem.

The leverage has moved upstream: defining constraints, surfacing tradeoffs and asking the right questions. In other words, prompting well.

  • Before AI

    Write the implementation

    Now

    Define the requirements precisely

  • Before AI

    Know every API

    Now

    Know what to ask for

  • Before AI

    Debug line by line

    Now

    Narrow the problem space quickly

How it works

  1. Step 1

    Pick a vague problem

    Every problem is deliberately under-specified, just like real engineering work.

  2. Step 2

    Write a prompt chain

    Ask clarifying questions, set constraints and explore tradeoffs. The AI answers as well as you ask.

  3. Step 3

    Get scored on your prompts

    Your prompts are scored, not the AI’s answers. Sharper questions earn higher scores.

One problem, three ways in.

“Design a URL shortener” has no single right opening. Here is how the first prompt changes with what you prioritise.

See real attempts
Security first
First thing I’d lock down is auth. Who can shorten a URL, and can they see analytics for links they didn’t create? Permissions need to be airtight before we talk scale.
Scale first
At 10K redirects per second the bottleneck is the read path. I’d assume an edge cache in front of everything and work backwards from there. The write path barely matters.
Users first
Before any infra: what does a broken redirect cost us? If users hit a dead link once, they leave. I’d rather over-invest in reliability than chase p99 latency on day one.

Find out how good your prompts really are.

Pick a problem, write your prompts and see how you compare with everyone who tried it before you.

Try a problemSee top attempts